KAIST Develops AI That Reconstructs Cloud-Covered Ground in Satellite Images with Photo-Like Clarity (IMAGE)

The Korea Advanced Institute of Science and Technology (KAIST)

KAIST Develops AI That Reconstructs Cloud-Covered Ground in Satellite Images with Photo-Like Clarity

Caption

From pretraining to task-specific applications. The left panel shows pretraining with satellite images, descriptive sentences, and location and GSD information. The center panel shows the resulting core image-generating model with 9 billion parameters. The right panel shows the model being fine-tuned separately for cloud removal and SAR-to-optical image translation. During pretraining, an AI specialized in satellite imagery provides a reference that helps the model learn image structure and meaning.

Credit

KAIST

Usage Restrictions

No

License

Licensed content

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.